None defined yet.
We are experimenting to see the effects of dataset composition (datacomp) on LLMs.
Docker for https://github.com/lee-ny/teaching_arithmetic
Setting up a MWE docker for training-arithmetic